Detecting Gesture Motions Through Smoke Or Glass

Table of Contents

Introduction

The objective of this experiment is to demonstrate the ability to detect gesture motions using a TI mmWave Sensor EVM through two impediments, glass and smoke. Two types of gesture motions were tested, a finger twirl motion and hand double swipe motion. Gestures were performed both without any impediment and then with the impediment introduced in between the sensor and the gestures.

Setup

The setup for this experiment can be divided into two parts, namely physical setup and software setup. These are detailed below:

Physical Setup

The physical setup includes information about the scene such as the distance between the target and the sensor antenna, the type of target used, hardware setup and any other physical considerations e.g. non-reflective surfaces or use of anechoic chamber etc. The physical setup used for this experiment is given below:

The following cases were tested for each of the two gestures:

Glass Impediment Smoke Impediment
Control Glass Sliding door open Fog machine off
Test Case Glass sliding door closed Fog machine on and producing smoke

Glass Impediment Setup

Smoke Impediment Setup

Software setup

Method

The steps presented below were repeated for each gesture and impediment combination.

  1. The physical and software setup was carried out as detailed above.
  2. The gesture was repeated at least 5x under the control case, with a few seconds in between each gesture.
  3. Impediment was introduced
    • For glass: The glass door was close as detailed above in the Physical Setup section.
    • For smoke: The fog generator was turned on as detailed above in the Physical Setup section.
  4. The gesture was repeated at least 5x with the impediment, with a few seconds in between each gesture.

Results

The steps presented above were repeated for each gesture and impediment type. The results are summarized in the following table:

Gesture Control With Impediment
Glass Double Swipe Detected Detected
Twirl Detected Detected
Smoke Double Swipe Detected Detected
Twirl Detected Detected

The following table shows simulatenous recordings of the gesture as performed and the response of the gesture recognition GUI:

Gesture Control With Glass Door Impediment
Double Swipe
Twirl
Gesture Control With Smoke Impediment
Double Swipe
Twirl

Conclusion

The experiment demonstrates that an IWR1443 mmWave sensor EVM is capable of gesture recognition with either glass or smoke as impediments between the sensor and the target hand making the gesture. The impediments do degrade the signals used for gesture detection. However, the degree to which this affects gesture recognition capabilities can be tuned by considering sensor parameters as well gesture recognition algorithm parameters.